基于分形维数的垩白米图像检测方法

Detection of Chalky Rice Based on Fractal Dimension

提出了一种基于分形维数的垩白米检测算法,并与基于垩白大小的检测算法进行了试验对比分析。分形维数包含了大米垩白区域的累计和空间分布特征,更能客观反映垩白区域的信息。试验结果表明,该算法的识别正确率为95.1%,可以有效识别垩白米,而且识别效果好于基于垩白大小的检测算法。

A new method, we named it F-method, based on the fractal dimension is presented in this paper. The F-method was used for detection of chalky rice and the result was compared with that detected by the C-method based on the chalkiness extent of rice kernels. The detection by the fractal dimension depends on more parameters, including the characteristics of chalkiness zones accumulation and space distribution, than that on only the rice kernel chalkiness. The experimental results showed that the F-method was more valid and efficient and its recognition accuracy reached to 95.1% that was higher than that from the C-method.